Autonomous Robot Driving using Sensor Fusion

In the rapidly evolving landscape of automation, robots and autonomous vehicles have become essential tools, driving innovation, improving efficiency and reliability, while integrating and cooperating with humans. This thesis presents the development of an Autonomous Mobile Robot (AMR) system based on the Yahboom ROSMASTER X3, from the assembly phase to code implementation. The system is powered by a NVIDIA Jetson Nano and actuated by a STM32-based board. The robot is equipped with a depth camera and a LiDAR sensor. The robot primary function is to track a moving target in real time while autonomously avoiding obstacles. The person detection is based on a previous thesis project on neural networks and real-time object tracking. However, the main focus of this project is the sensor fusion of the camera and LiDAR in ROS. Additionally, the thesis explores the architecture of the ROSMASTER X3, the use of Docker and containers, and the communication protocols implemented. Various tests were conducted to assess the system performance in complex environments and under different conditions, focusing on real-time response, obstacle avoidance accuracy, and smoothness in movement transitions. The results indicate that ROS2 architecture and the integration of sensor fusion techniques significantly enhance the robot autonomous capabilities, making it suitable for dynamic environments. This work contributes to the wider field of autonomous mobile robotics by demonstrating an effective implementation of ROS2-based systems for mobile robots.